Fermion-number violation in regularizations that preserve fermion-number symmetry

TL;DR

This paper investigates how fermion-number violating effects, specifically 't Hooft vertices, are realized in regularizations that preserve fermion number symmetry, resolving apparent conflicts between symmetries at different levels.

Contribution

It demonstrates that in regularizations preserving fermion number, the chiral U(1) symmetry is spontaneously broken, and 't Hooft vertices are correctly recovered through a careful limiting procedure.

Findings

't Hooft vertices are recovered in the continuum limit.

Spontaneous breaking of chiral U(1) symmetry occurs in the covariantly gauge-fixed theory.

The approach resolves the paradox between symmetry preservation and non-perturbative effects.

Abstract

There exist both continuum and lattice regularizations of gauge theories with fermions which preserve chiral U(1) invariance ("fermion number"). Such regularizations necessarily break gauge invariance but, in a covariant gauge, one recovers gauge invariance to all orders in perturbation theory by including suitable counterterms. At the non-perturbative level, an apparent conflict then arises between the chiral U(1) symmetry of the regularized theory and the existence of 't Hooft vertices in the renormalized theory. The only possible resolution of the paradox is that the chiral U(1) symmetry is broken spontaneously in the enlarged Hilbert space of the covariantly gauge-fixed theory. The corresponding Goldstone boson is unphysical. The theory must therefore be defined by introducing a small fermion-mass term that breaks explicitly the chiral U(1) invariance, and is sent to zero after the infinite-volume limit has been taken. Using this careful definition (and a lattice regularization) for the calculation of correlation functions in the one-instanton sector, we show that the 't Hooft vertices are recovered as expected.